← All changes | includes/class-mxchat-utils.php +23 -40 3.2.2 → 3.1.4 View file →

		@@ -177,15 +177,9 @@
177	177	}
178	178
179	179	// ===== FIXED: Generate unique identifier for manual content =====
180	180	$original_source_url = $source_url;
181		- // Check if this is truly manual content (no URL at all) vs a real URL that filter_var rejects
182		- // filter_var(FILTER_VALIDATE_URL) rejects valid URLs with encoded chars, non-ASCII, fragments, etc.
183		- // Use a looser check: if it starts with http(s):// or has a scheme, it's a URL
184		- $has_url_scheme = !empty($source_url) && preg_match('#^https?://#i', $source_url);
185		- // Treat legacy mxchat.ai source URLs as manual — old bug assigned the site URL to manual entries
186		- $is_legacy_mxchat_url = $has_url_scheme && strpos($source_url, 'mxchat.ai') !== false;
187		- $is_manual_content = empty($source_url) || $source_url === '' || !$has_url_scheme || $is_legacy_mxchat_url;
	181	+ $is_manual_content = empty($source_url) || $source_url === '' || !filter_var($source_url, FILTER_VALIDATE_URL);
188	182
189	183	if ($is_manual_content) {
190	184	// Generate unique identifier for manual content to prevent overwrites
191	185	$source_url = 'mxchat://manual-content/' . time() . '-' . wp_generate_password(8, false);
		@@ -309,14 +303,14 @@
309	303	// ===== UPDATED: Handle manual content with unique vector IDs =====
310	304	if ($vector_id) {
311	305	// Use provided vector ID
312	306	//error_log('[MXCHAT-PINECONE-MAIN] Using provided vector ID: ' . $vector_id);
313		- } elseif (!empty($url) && preg_match('#^https?://#i', $url)) {
314		- // For URLs, use URL-based ID (existing behavior)
	307	+ } elseif (!empty($url) && filter_var($url, FILTER_VALIDATE_URL)) {
	308	+ // For valid URLs, use URL-based ID (existing behavior)
315	309	$vector_id = md5($url);
316	310	//error_log('[MXCHAT-PINECONE-MAIN] Generated vector ID from URL: ' . $vector_id);
317	311	} else {
318		- // For manual content (empty/no URL scheme), generate unique ID
	312	+ // For manual content (empty/invalid URL), generate unique ID
319	313	$vector_id = 'manual_' . time() . '_' . substr(md5($content . microtime(true)), 0, 8);
320	314	//error_log('[MXCHAT-PINECONE-MAIN] Generated unique vector ID for manual content: ' . $vector_id);
321	315	}
322	316	// ===== END UPDATE =====
		@@ -352,9 +346,9 @@
352	346	// Fallback to old detection logic for backwards compatibility
353	347	$is_product = false;
354	348	$content_type = 'manual'; // Default for manual content
355	349
356		- if (!empty($url) && preg_match('#^https?://#i', $url)) {
	350	+ if (!empty($url) && filter_var($url, FILTER_VALIDATE_URL)) {
357	351	$is_product = (strpos($url, '/product/') !== false || strpos($url, '/shop/') !== false);
358	352	$content_type = $is_product ? 'product' : 'content';
359	353	}
360	354	}
		@@ -804,34 +798,31 @@
804	798
805	799	// Add the original single-vector ID (for non-chunked content)
806	800	$vectors_to_delete[] = $base_vector_id;
807	801
808		- // Pinecone /vectors/list is a GET endpoint with query-string parameters; a POST here returns a
809		- // non-200 silently and we end up only deleting the base vector, leaving chunks orphaned.
810		- $query_params = array(
	802	+ // Use Pinecone list API to find all chunk vectors with this prefix
	803	+ $list_url = "https://{$host}/vectors/list";
	804	+
	805	+ $list_body = array(
811	806	'prefix' => $base_vector_id . '_chunk_',
812		- 'limit' => 100,
	807	+ 'limit' => 100
813	808	);
	809	+
814	810	if (!empty($namespace)) {
815		- $query_params['namespace'] = $namespace;
	811	+ $list_body['namespace'] = $namespace;
816	812	}
817	813
818		- $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params);
	814	+ $list_response = wp_remote_post($list_url, array(
	815	+ 'headers' => array(
	816	+ 'Api-Key' => $api_key,
	817	+ 'accept' => 'application/json',
	818	+ 'content-type' => 'application/json'
	819	+ ),
	820	+ 'body' => wp_json_encode($list_body),
	821	+ 'timeout' => 30
	822	+ ));
819	823
820		- // Paginate in case a URL has more than 100 chunks.
821		- do {
822		- $list_response = wp_remote_get($list_url, array(
823		- 'headers' => array(
824		- 'Api-Key' => $api_key,
825		- 'accept' => 'application/json',
826		- ),
827		- 'timeout' => 30,
828		- ));
829		-
830		- if (is_wp_error($list_response) || wp_remote_retrieve_response_code($list_response) !== 200) {
831		- break;
832		- }
833		-
	824	+ if (!is_wp_error($list_response)) {
834	825	$list_data = json_decode(wp_remote_retrieve_body($list_response), true);
835	826	if (!empty($list_data['vectors'])) {
836	827	foreach ($list_data['vectors'] as $vector) {
837	828	if (isset($vector['id'])) {
		@@ -838,17 +829,9 @@
838	829	$vectors_to_delete[] = $vector['id'];
839	830	}
840	831	}
841	832	}
842		-
843		- $next_token = $list_data['pagination']['next'] ?? '';
844		- if (empty($next_token)) {
845		- break;
846		- }
847		-
848		- $query_params['paginationToken'] = $next_token;
849		- $list_url = "https://{$host}/vectors/list?" . http_build_query($query_params);
850		- } while (true);
	833	+ }
851	834
852	835	if (empty($vectors_to_delete)) {
853	836	//error_log('[MXCHAT-CHUNK-DELETE] No vectors found to delete');
854	837	return true;